NEW YORK (AP) _ In a new sign of how slippery a foe the AIDS virus is, researchers reported today that it can infect key blood cells even after it's trapped and chemically handcuffed by the body.

The result shows that researchers must find ways to attack the trapped HIV, researcher Gregory Burton said.

Scientists have long known that large populations of HIV become trapped by a weblike mesh in lymphoid tissues like the tonsils, spleen and lymph nodes.

Studies show that HIV infects blood cells in lymphoid tissues, but it hasn't been clear whether the trapped HIV was responsible.

That's because the trapped viruses are ``handcuffed,'' covered with antibodies and other immune system proteins that should prevent them from infecting cells.

But the new study says they are infectious, and it blames the cells that form the weblike mesh. Somehow, these cells let trapped HIV continue to infect despite its handcuffs, researchers from Virginia Commonwealth University in Richmond report in today's issue of the journal Nature.

Burton, an assistant professor of microbiology and immunology, said current anti-HIV drugs may not work against the trapped virus because they are aimed at blocking reproduction, and HIV does not reproduce while trapped.

``We go out and destroy the virus at other sites, but it remains sitting (while trapped) ... waiting to infect as soon as it gets an opportunity,'' he said.

The web-making cells are called follicular dendritic cells. Their normal job is to trap and hold bits of foreign material, which serve to remind the disease-fighting immune system what germs and other invaders look like.

The new work was done in mice and in test tubes. Researchers found that when HIV was handcuffed by so-called neutralizing antibodies, it could still infect its key target, T cells of the immune system, but only if dendritic cells were present.

That happened even when the levels of neutralizing antibodies were a million times greater than what is normally needed to block infection, Burton said. ``You really wouldn't think it had any chance at all of being able to infect, and yet it clearly does,'' he said.

It's not clear how dendritic cells permit HIV infection. But since their normal job is to show bits of germs to the immune system, they may expose the AIDS virus in a way that lets it infect nearby T cells, Burton said.

One possibility for treatment might be to find ways to keep HIV from sticking to dendritic cells, so that it could not longer get their help, Burton said.

Dr. Ashley Haase of the University of Minnesota, who has studied the behavior of HIV in lymphoid tissue, said the ability of dendritic cells to make the trapped virus infectious is ``quite extraordinary.''

The result shows you have ``an enormous pot of infectious virus there'' that will have to be eradicated by new drugs that can attack the trapped HIV, he said.